1. Technical Field
The invention concerns plasma reactors used in processing workpieces in the manufacturing of items such as microelectronic circuits, flat panel displays and the like, and in particular plasma sources therefor.
2. Background Art
The trend in microelectronic circuits toward ever increasing densities and smaller feature sizes continues to make plasma processing of such devices more difficult. For example, the diameter of contact holes has been reduced while the hole depth has increased. During plasma-enhanced etching of a dielectric film on a silicon wafer, for example, the etch selectivity of the dielectric material (e.g. silicon dioxide) to photoresist must be sufficient to allow the etch process to etch a contact hole whose diameter is ten to fifteen times its depth, without appreciably disturbing the photoresist mask defining the hole. This task is made even more difficult because the recent trend toward shorter wavelength light for finer photolithography requires a thinner photoresist layer, so that the dielectric-to-photoresist etch selectivity must be greater than ever. This requirement is more readily met using processes having relatively low etch rates, such as dielectric etch processes employing a capacitively coupled plasma. The plasma density of a capacitively coupled plasma is relatively less than that of an inductively coupled plasma, and the capacitively coupled plasma etch process exhibits good dielectric-to-photoresist etch selectivity. The problem with the capacitively coupled process is that it is slow and therefore relatively less productive. Another problem that arises in such etch processes is non-uniform plasma distribution.
In order to increase productivity or etch rate, higher density plasmas have been employed. Typically, the high density plasma is an inductively coupled plasma. However, the process precursor gases tend to dissociate more rapidly in such a high density plasma, creating a higher plasma content of free fluorine, a species which reduces the etch selectivity to photoresist. To reduce this tendency, fluoro-carbon process gases such as CF2 are employed which dissociate in a plasma into fluorine-containing etchant species and one or more polymer species which tend to accumulate on non-oxide containing surfaces such as photoresist. This tends to increase etch selectivity. The oxygen in the oxygen-containing dielectric material promotes the pyrolization of the polymer over the dielectric, so that the polymer is removed, allowing the dielectric material to be etched while the non-oxygen containing material (e.g., the photoresist) continues to be covered by the polymer and therefore protected from the etchant. The problem is that the increase in contact opening depth and decrease in photoresist thickness to accommodate more advanced device designs has rendered the high density plasma process more likely to harm the photoresist layer during dielectric etching. As the plasma density is increased to improve etch rate, a more polymer-rich plasma must be used to protect the non-oxygen containing material such as photoresist, so that the rate of polymer removal from the oxygen-containing dielectric surfaces slows appreciably, particularly in small confined regions such as the bottom of a narrow contact opening. The result is that, while the photoresist may be adequately protected, the possibility is increased for the etch process to be blocked by polymer accumulation once a contact opening reaches a certain depth. Typically, the etch stop depth is less than the required depth of the contact opening so that the device fails. The contact opening may provide connection between an upper polysilicon conductor layer and a lower polysilicon conductor layer through an intermediate insulating silicon dioxide layer. Device failure occurs, for example, where the etch stop depth is less than the distance between the upper and lower polysilicon layers. Alternatively, the possibility arises of the process window for achieving a high density plasma without etch stop becoming too narrow for practical or reliable application to the more advanced device designs such as those having contact openings with aspect ratios of 10:1 or 15:1.
What would be desirable at present is a reactor that has the etch rate of an inductively coupled plasma reactor (having a high density plasma) with the selectivity of a capacitively coupled reactor. It has been difficult to realize the advantages of both types of reactors in a single machine led reactor.
One problem with high density inductively coupled plasma reactors, particularly of the type having an overhead coil antenna facing the wafer or workpiece, is that as the power applied to the coil antenna is increased to enhance the etch rate, the wafer-to-ceiling gap must be sufficiently large so that the power is absorbed in the plasma region well above the wafer. This avoids a risk of device damage on the wafer due to strong RF fields. Moreover, for high levels of RF power applied to the overhead coil antenna, the wafer-to-ceiling gap must be relatively large, and therefore the benefits of a small gap cannot be realized.
If the ceiling is a semiconductive window for the RF field of an inductively coupled reactor, or a conductive electrode of a capacitively coupled reactor, then one benefit of a small wafer-to-ceiling gap is an enhanced electric potential or ground reference that the ceiling could provide across the plane of the wafer at a relatively small gap distance (e.g., on the order of 1 or 2 inches).
Therefore, it would be desireable to have a reactor not only having the selectivity of a capacitively coupled reactor with the ion density and etch rate of an inductively coupled reactor, but further having none of the conventional limitations on the wafer-to-ceiling gap length other than a fundamental limit, such as the plasma sheath thickness, for example. It would further be desireable to have a reactor having the selectivity of a capacitively coupled reactor and the etch rate of an inductively coupled reactor in which the ion density and etch rate can be enhanced without necessarily increasing the applied RF plasma source power.